KAREN C. GAREIS AND ROSALIND CHAIT BARNETT
Brandeis University
ROBERT T. BRENNAN
Harvard University*
l
Individual and Crossover Effects of Work Schedule Fit:
A Within-Couple Analysis
A couple-level analysis with a sample of 105 female reduced-hours physicians and their fulltime–employed husbands found individual and
spouse crossover effects: Each spouse’s ratings of
own schedule fit predict own job-role quality;
wives’ ratings of partner/family schedule fit predict their marital-role quality, with a similar trend
for husbands; husbands’ ratings of own schedule
fit predict wives’ marital-role quality; and husbands’ ratings of partner/family schedule fit and
wives’ ratings of own schedule fit predict husbands’ psychological distress. These findings
highlight the interdependence of couple members’
experiences and illustrate potential costs of wives’
trading off time at work for time at home.
The literature on the work-family interface has
been characterized by a focus on employees as
autonomous agents, not as members of dyads in
which each partner’s job and family experiences
affect the other partner’s social-role experiences.
A number of studies provide convincing evidence
Brandeis University, Women’s Studies Research Center,
Mailstop 079, 515 South Street, Waltham, MA 02454-9110
(gareis@brandeis.edu).
*Harvard University, Graduate School of Education, Monroe C. Gutman Library, Appian Way, Cambridge, MA
02138-3704.
Key Words: dual-earner couples, reduced work schedules,
schedule fit, within-couple analysis, work-family tradeoffs.
of crossover effects from one member of a couple
to the other, however (e.g., Barling, 1984; Hammer, Allen, & Grigsby, 1997; Ozer, Barnett, Brennan, & Sperling, 1998; Raudenbush, Brennan, &
Barnett, 1995; Rook, Dooley, & Catalano, 1991).
To adequately model the work-family interface,
one must take into account the relationship between one partner’s job characteristics and outcomes for the other partner; ideally, such data
should be collected from both partners in order to
allow for the analysis of crossover effects.
In the present study, we have data from both
members of married couples and employ an analysis strategy, hierarchical linear modeling (HLM;
Raudenbush & Bryk, 2002), that retains the couple membership of each spouse (Raudenbush et
al., 1995). A common shortcoming of several previous studies of dual-earner couples is that they
estimate separate predictive models for husbands
and wives (e.g., Westman & Etzion, 1995), in essence discarding the relationship between the two
and treating them as if they came from independent populations. HLM is also a flexible strategy
to test for crossover effects in which a predictor
variable measured in one spouse is associated with
an outcome variable measured in the other spouse
(Barnett & Brennan, 1998; Raudenbush et al.).
Applying HLM to the analysis of data from couples accounts for the dependence of observations
for each of the members, controls for both individual- and couple-level predictors, adjusts for
Journal of Marriage and Family 65 (November 2003): 1041–1054
1041
1042
measurement error in estimating within-couple
correlations and in estimating proportions of variance explained by the model, and allows for explicit testing of gender differences (Barnett, Brennan, Raudenbush, Pleck, & Marshall, 1995;
Barnett, Marshall, Raudenbush, & Brennan, 1993;
Raudenbush et al.). Finally, the same modeling
strategy is suitable for multiple outcomes for each
person (Brennan, Kim, Wenz-Gross, & Siperstein,
2001; Supovitz & Brennan, 1997). Thus, our approach allows us to model simultaneously six outcomes (husband’s and wife’s job-role quality, marital-role quality, and psychological distress) in a
single, two-level HLM (Ozer et al., 1998).
In addition to a narrow focus on employees as
individuals, the mainstream organizational psychology literature on work-family issues typically
characterizes work and family as spheres of life
that are in competition for such scarce resources
as time and attention (Barnett, 1998; Gutek, Searle, & Klepa, 1991). Excessive work hours have
been related to such negative outcomes as marital
tension (Hughes & Galinsky, 1994); numerous
studies challenge the zero-sum assumption, however, finding either no relationship (e.g., Crouter,
Bumpus, Head, & McHale, 2001) or a positive
relationship (e.g., Ozer et al., 1998; Wethington &
Kessler, 1989) between number of hours worked
and outcomes. Marks, Huston, Johnson, and
MacDermid (2001) found a positive relationship
between work hours and role balance among
wives, but a negative relationship between the
same variables among husbands in their study. Inconsistent findings about the relationship between
work hours and outcomes, along with a growing
body of research demonstrating work-family synergy (Bailyn, 1993) and salutary effects of multiple-role involvement (Barnett & Hyde, 2001;
Marks & MacDermid, 1996) also challenge the
inevitability of the zero-sum assumption.
Work and family may more accurately be depicted as overlapping spheres that are often in harmony (see Barnett, 1998, for a fuller discussion).
The concept of fit, first introduced by Pittman
(1994; see also Bowen, 1998; Bowen, Orthner, &
Bell, 1997), comports with this approach. It is assumed that dual-earner couples formulate adaptive
strategies for maximizing both spouses’ ability to
meet workplace and family needs (Moen & Wethington, 1992), and that when workers are able to
realize their strategies, they experience compatibility and low distress. We focus on schedule fit
because a key aspect of such strategies is the number, distribution, and flexibility of each partner’s
Journal of Marriage and Family
work hours, although we recognize that fit could
also be assessed with respect to such dimensions
as energy, strain, and behaviors (MacDermid et
al., 2000).
We operationalize schedule fit as respondents’
subjective assessment of the degree to which their
own and their spouses’ work arrangements meet
their needs and those of their spouses and other
dependents (Barnett, Gareis, & Brennan, 1999),
reflecting the fact that schedule fit is evaluated in
light of the needs of all family members. We distinguish between two aspects of schedule fit: the
fit of one’s own work schedule for oneself (self/
self schedule fit) and the fit of one’s spouse’s work
schedule for oneself and other family members
(partner/family schedule fit). Although we also
have data on self/family schedule fit, we choose
to use spouse ratings of partner/family schedule
fit instead (see Measures for further details). It is
expected that one’s own and one’s spouse’s ratings
of self/self and partner/family schedule fit will be
associated with outcomes such as job-role quality,
marital-role quality, and psychological distress.
In the present sample of dual-earner couples,
the wives are physicians who have voluntarily reduced their work hours to better manage work and
family demands, whereas their husbands work
full-time schedules. This sample is of interest for
several reasons. First, dual-earner couples are becoming more common: Between 1970 and 1993,
dual-earner couples increased from 39% to 61%
of all married couples (Blau, Ferber, & Winkler,
1998). Further, in such couples, women are more
likely than are men to adapt their work schedules
to meet family needs (see Spain & Bianchi, 1996,
for a review), as have the wives in our sample.
Finally, Carr (2002) argues that women who have
invested heavily in their education and careers, as
have the physicians in this sample, will be more
distressed if they cut back on their paid employment to meet family responsibilities than will less
career-oriented women, because such labor market
reductions may conflict with their expectations for
achieving occupational success.
Several recent studies have examined the effects on individuals of adjusting work schedules
to accommodate family needs. Findings have been
somewhat mixed. For example, in a sample of
dual-earner families with both child-care and elder-care responsibilities, Brockwood, Hammer,
Neal, and Colton (2002) found that for wives,
making work accommodations such as schedule
changes was associated with greater family satisfaction, but also with lower job satisfaction; thus,
Work Schedule Fit
FIGURE 1. INDIVIDUAL
1043
AND
CROSSOVER EFFECTS
work accommodations seemed to represent a
tradeoff for these women. Milkie and Peltola
(1999) found that for married, employed men and
women, the total number of work sacrifices made
for family reasons was negatively related to sense
of work-family balance. It is difficult to interpret
this finding, however, because the authors defined
work sacrifices as including not only refusing a
promotion, refusing overtime, and cutting hours,
but also as taking on more paid work for family
reasons, a strategy much more common among
men in their sample. Carr (2002) studied the effects of work adjustments made to meet childrearing needs—including stopping work, cutting
back on work, and changing jobs—on evaluations
of work opportunities and self-acceptance in men
and women from different birth cohorts. She
found that work adjustments did affect outcomes
depending on gender and cohort, but significant
effects were associated with stopping work rather
than with a variable combining cutting back on
work (as wives in the present sample did) and
changing jobs.
None of the studies described above examined
the effects of work schedule adjustments at the
couple level, analyzing husband and wife outcomes jointly and allowing for crossover effects
from one spouse to the other. The present study is
responsive to Carr’s (2002) suggestions that future
research should explore couple-level as well as
individual-level work-family strategies and examine additional occupational and marital outcomes.
We test the six hypotheses described below.
For each pair of hypotheses, the first predicts an
individual effect in which that spouse’s own
schedule fit ratings predict that spouse’s own outcomes (see Figure 1, solid lines), and the second
OF
SPOUSES’ RATINGS
OF
WORK SCHEDULE FIT
predicts a crossover effect in which one spouse’s
schedule fit ratings predict the other spouse’s outcomes (see Figure 1, dotted lines).
1. a. Job-role quality will be predicted by own
ratings of self/self schedule fit for both husbands and wives. That is, to the extent that
one feels that one’s own work schedule fits
one’s own needs, one’s own job-role quality
will also be positive.
b. Job-role quality will be predicted by spouse
ratings of partner/family schedule fit in a
crossover effect for both husbands and
wives. That is, to the extent that one’s
spouse feels that one’s own work schedule
fits family needs, one’s own job-role quality
will also be positive.
2. a. Marital-role quality will be predicted by
own ratings of self/self schedule fit and own
ratings of partner/family schedule fit for
both husbands and wives. That is, to the extent that one feels that one’s own work
schedule fits one’s own needs and that one’s
partner’s work schedule fits family needs,
one’s own marital-role quality will also be
positive.
b. Marital-role quality will be predicted by
spouse ratings of partner/family schedule fit
and spouse ratings of self/self schedule fit
in crossover effects for both husbands and
wives. That is, to the extent that one’s
spouse feels that one’s own work schedule
fits family needs and that the spouse’s own
work schedule fits the spouse’s needs, one’s
own marital-role quality will also be positive.
3. a. Psychological distress will be predicted by
own ratings of self/self and partner/family
1044
Journal of Marriage and Family
schedule fit for both husbands and wives.
That is, to the extent that one feels that one’s
own work schedule fits one’s own needs and
that one’s partner’s work schedule fits family needs, one’s own distress will be low.
b. Psychological distress will be predicted by
spouse’s ratings of self/self and partner/family schedule fit in crossover effects for both
husbands and wives. That is, to the extent
that one’s spouse feels that the spouse’s
work schedule fits the spouse’s needs and
that one’s own work schedule fits family
needs, one’s own psychological distress will
be low.
In all analyses, we control for known correlates
of job-role quality, marital-role quality, and psychological distress, including both spouses’ negative affectivity and the couple’s income, marriage
length, and number of children. We control for
each spouse’s negative affectivity, a mood-dispositional trait to view the world negatively that is
thought to account for spuriously high correlations
between self-report measures of predictors and
outcomes, especially in cross-sectional analyses
(Brennan & Barnett, 1998; Burke, Brief, &
George, 1993). Because reduced hours are usually
only an option for those workers whose partners
earn enough money to offset the consequent reduced income, and because income has been
linked to a number of physical and psychological
well-being outcomes, we also control for household income.
Length of marriage and number of children
have been related to marital-role quality, and number of children might also affect schedule fit because parents of more children might have more
difficulty achieving good schedule fit. Therefore,
we control for length of marriage and number of
children. Finally, because this is a study of work
schedules, we include both spouses’ work hours
as additional predictors.
METHOD
Sample
As part of a larger study of the anticipated and
unanticipated consequences of reduced-hours
work on physicians, their spouses, and their employers (e.g., Barnett et al., 1999), we interviewed
105 female, Boston-area physicians who reported
voluntarily working reduced-hours schedules and
their full-time–employed husbands. Reduced
hours was defined as working a schedule that the
physician considered to be reduced for at least the
preceding 3 months. The women had adjusted
their work schedules in order to better manage
their work and family demands, as demonstrated
by their responses to a 16-item measure of their
reasons for reducing their work hours, which was
based on content analyses of preliminary openended interviews with 24 reduced-hours physicians and their spouses. By far the most common
reason for reducing work hours, endorsed as at
least somewhat important by 97.0% of the present
sample, was ‘‘need for more time with my children.’’ No other reason was endorsed by more
than 66.7% of respondents, although a plurality of
respondents endorsed two items about wanting to
spend more time with their spouses, two items
about needing more personal time, one item about
needing to reduce work-related pressure, and one
item about not wanting to work traditional fulltime physicians’ hours.
The majority (92.4%) of the reduced-hours
physicians were White, as were their husbands
(97.1%); on average, they were 41.4 years of age
(SD 5 5.4), whereas their husbands were 42.9
(SD 5 6.7). On the average, reduced-hours physicians had been working their current reduced
schedules for about 4 years (M 5 47.4 months,
SD 5 45.0), whereas their husbands had been
working their schedules for almost 7 years (M 5
80.6 months, SD 5 64.3). On average, the reduced-hours physicians worked 29.2 hours per
week (Mdn 5 30, SD 5 8.4), although it should
be noted that the women’s reduced-hours schedules ranged up to 50 hours per week. In contrast,
their husbands averaged 52.2 hours per week
(Mdn 5 50, SD 5 9.9), ranging up to 75 hours
per week. Although the wives voluntarily reduced
their hours, only 53.3% reported working their
preferred number of hours; 39.0% reported working more and 7.6% reporting working fewer than
they would prefer. Among the husbands, only
39.0% reported working their preferred number of
hours, with 57.1% reporting working more and
3.8% reporting working fewer than they would
prefer.
The majority of the reduced-hours physicians
(62.1%) were married to other physicians. On average, the couples had been married 12.4 years
(SD 5 6.5). Of the men, 84.8% were in their first
marriage, as were 91.4% of the women. The vast
majority of the couples (96.2%) were parents; of
those, the average number of children was 2.3 (SD
5 0.9), and the average age of the children was
Work Schedule Fit
8.2 years (SD 5 6.4). The median yearly household income was $200,000, although the range
was broad (SD 5 $120,426).
We used several strategies to develop the sample, including random sampling from the registry
of the Massachusetts Board of Certification in
Medicine; requesting nominations from respondents; and contacting area HMOs, hospitals, and
practice partnerships to ask for assistance in recruitment. Our response rates (35.5% among randomly sampled respondents and 43.0% among
nominated respondents) compare favorably with
those achieved in other studies of physicians (Parsons, Warnecke, Czaja, Barnsley, & Kaluzny,
1994; Sobal et al., 1990), especially in light of the
fact that both the reduced-hours physicians and
their husbands had to agree to participate.
Procedures
Each respondent participated in a private, face-toface, 60-minute quantitative survey. Respondents
also completed a 20-minute mailed survey to be
returned at the time of the interview. The interview and survey together comprised about 558
items covering respondents’ work arrangements;
social-role quality (as spouse, employee, and parent); and quality of life.
Measures
Schedule fit was assessed using two subscales derived from a 9-item scale designed to measure
subjective appraisals of the degree to which both
spouses’ work arrangements, in terms of number
and distribution of work hours, meet their own
needs and those of other family members (Barnett
et al., 1999). Exploratory factor analysis results
showed that scale items loaded onto three factors
representing the fit of one’s own schedule for oneself (self/self schedule fit), the fit of one’s own
schedule for other family members (self/family
schedule fit), and the fit of one’s spouse’s schedule
for other family members (partner/family schedule
fit).
Because we have ratings from both members
of each couple, one spouse’s ratings of self/family
schedule fit are conceptually redundant with the
other spouse’s rating of partner/family schedule
fit. In other words, we have ratings of how the
wife’s schedule works for family members from
both the wife (self/family schedule fit) and the
husband (partner/family schedule fit); in an analogous fashion, we have two sources for how the
1045
husband’s schedule works for other family members. These ratings showed only low to moderate
correlations between partners, with a correlation
coefficient of 0.44 (p 5 .000) between husband
and wife ratings of how the husband’s schedule
works for family members, and a correlation coefficient of 0.25 (p 5 .010) between husband and
wife ratings of how the wife’s schedule works for
family members. We judged that the partner/family schedule fit rating was less biased because it
comes from one of the affected parties; that is, a
family member rates how the partner’s schedule
fits family needs.
The self/self schedule fit subscale consists of
four items such as ‘‘Taking into account your current work hours and schedule, how well is your
work arrangement working for you?’’ The partner/
family schedule fit subscale consists of three items
such as ‘‘Taking into account your spouse’s current work hours and schedule, how well is your
arrangement working for your child(ren)?’’ Internal consistency for self/self schedule fit and partner/family schedule fit are good, with Cronbach’s
a 5 .87 and .78, respectively.
Overall, the reduced-hours physicians reported
fairly good self/self schedule fit, with an average
rating of 5.4 (SD 5 1.1, range 5 2.5–7.0) on a
scale from 1 to 7, corresponding to a point between slightly well and mostly well. Their husbands gave similar ratings of self/self schedule fit,
with an average rating of 5.3 (SD 5 1.2, range 5
1.0–7.0). In contrast, the reduced-hours physicians rated partner/family schedule fit lower, with
an average rating of 4.7 (SD 5 1.3, range 5 2.0–
7.0). This was significantly lower, t(201.7) 5
2.94, p 5 .004, than their husbands’ ratings of
partner/family schedule fit, which averaged 5.2
(SD 5 1.1, range 5 2.7–7.0). It is not surprising
that the husbands rate the schedules of their reduced-hours wives more positively than the reduced-hours wives rate the schedules of their fulltime–employed husbands.
Job-role quality was assessed with a modified
version of a 28-item measure on which respondents were instructed to rate on a scale ranging
from 1 (not at all) to 4 (considerably) the extent
to which each item was rewarding or of concern
(Barnett & Brennan, 1995, 1997). Items covered
job conditions in the areas of skill discretion, decision authority, job demands, pay adequacy, job
security, and supervisor relations; three original
scale items on schedule control were omitted from
score computation in the present analysis to avoid
inflating the relationship between schedule fit and
1046
job-role quality. Concerns were negatively
weighted and rewards positively weighted in constructing the score, which was the weighted average of item scores (Barnett et al., 1993). Internal
consistency is good, with Cronbach’s a 5 .77 for
rewards and .81 for concerns in the present sample. Job-role quality was fairly high in this sample,
averaging 1.2 (SD 5 0.7) on a scale ranging from
23.0 to 13.0.
Marital-role quality was measured using a 15item brief form (Hyde & Plant, 1996) of the Marital-role Quality Scale (Barnett et al., 1993). As
with job-role quality, respondents indicated on a
4-point scale the degree to which each item was
rewarding or of concern. Internal consistency is
very good, with Cronbach’s a 5 .87 for rewards
and .89 for concerns in the present sample. Marital-role quality was high in this sample, averaging
2.0 (SD 5 1.0) on a scale ranging from 23.0 to
13.0.
Psychological distress was assessed by asking
respondents to indicate on a 5-point scale from 0
(not at all) to 4 (extremely) how often in the past
week they were bothered by each of 10 symptoms
of anxiety and 14 symptoms of depression (Derogatis, 1975). Anxiety and depression scores were
combined to create a measure of psychological
distress (Barnett et al., 1993). The combined score
has very good internal consistency, with Cronbach’s a 5 .91 in the present sample. Psychological distress was low in this sample, averaging
11.0 (SD 5 9.7) on a scale ranging from 0 to 96.
Work hours was assessed by asking respondents to estimate the number of hours worked in
an average work week, excluding on-call hours.
Household income per capita was calculated
by summing each spouse’s report of own income
and then dividing by the number of people in the
household. The distribution of this variable is
skewed, so we used the natural log of per capita
income in the analyses.
Number of children at home and number of
years married are self-explanatory.
Negative affectivity was assessed with the 10item Trait Anxiety Scale (Spielberger, 1983), on
which respondents indicated on a 4-point scale
from 1 (almost never) to 4 (almost always) how
characteristic the traits were of them. Test-retest
correlations ranged from .73 to .86 in college populations over a 2-year period (Spielberger) and
was .77 over a 1-year period in a sample of fulltime employed, dual-earner couples (Barnett et al.,
1995). Internal consistency is very good, with
Cronbach’s a 5 .83 and .85 (Brennan & Barnett,
Journal of Marriage and Family
1998) and .88 in the present sample. Negative affectivity was fairly low in this sample, averaging
20.6 (SD 5 5.8) on a scale ranging from 10 to
40.
Analytic Approach
Analyzing data from married couples requires a
technique that takes into account the shared variance of outcomes. In the past decade, several
methods have been proposed for handling such
data (Bray, 1995; Bray, Maxwell, & Cole, 1995;
Maguire, 1999). The use of HLM is a flexible
technique that takes into account the interdependence of the observations (Barnett et al., 1993;
Brennan, Barnett, & Gareis, 2001; Windle & Dumenci, 1997). Because it is possible that outcomes
within a social system may be predicted not just
by one’s own experiences, but also by the experiences of another member of the system (in this
case, the spouse), it is also critical that the analytic
technique be able to model crossover relationships; HLM is readily able to accommodate the
modeling of one couple member’s outcomes as a
function of the spouse’s experiences (Barnett &
Brennan, 1998; Raudenbush et al., 1995).
Earlier approaches to the modeling of couples
provided no method for testing whether a given
effect was greater or smaller for one member of
the couple, because available tests for the differences in regression coefficients and their analogues required the unsupportable assumption of
independent observations; in other words, that
quality-of-life outcomes such as marital-role quality or psychological distress are uncorrelated in
marital pairs. By modeling the outcomes for both
members of the couple simultaneously in a single
model, HLM allows for tests of the difference between effects without bias or mis-specification
(Raudenbush & Bryk, 2002; Raudenbush, Bryk,
Cheong, & Congdon, 2000).
In this analysis, we expand upon the original
adaptation of HLM proposed by Barnett et al.
(1993) for the analysis of data from married couples. Although the original model is in a strict
sense multivariate with respect to outcomes (Raudenbush et al., 1995), this fact is often overlooked
(e.g., Maguire, 1999) because both of the outcomes are of the same construct (e.g., psychological distress), albeit displayed as separate equations for the two couple members. Because it is
already multivariate, the original hierarchical
model for the analysis of data from dyads, in
which the outcomes specific to each member of a
Work Schedule Fit
couple are estimated simultaneously in a single
model, is easily expanded to accommodate multiple outcomes per respondent (Brennan, Kim, et
al., 2001; Supovitz & Brennan, 1997), also in a
single model. In particular, the multivariate outcome model was adapted to the study of married
couples with multiple outcomes for each member
of the couple by Ozer et al. (1998), who modeled
marital-role quality and psychological distress for
each member of a couple as latent outcomes in a
single model estimated as an HLM.
In the present analysis, we expand the model
to estimate simultaneously in one model six outcomes (husband’s and wife’s job-role quality, marital-role quality, and psychological distress). This
single-model approach takes into account both the
interrelatedness of the outcomes across members
of the couples as well as the within-individual correlations across the three measures. We retained
the use of parallel scales to estimate latent true
score outcomes for all three of the outcome measures.
We specified a priori a model (Model 4) in
which the effects of number of children, household income, marriage length, and both spouses’
negative affectivity were controlled, as were the
effects of both spouses’ work hours. Although this
was the only model of theoretical interest, in order
to deal with likely effects of multiple collinearity,
particularly as might affect the simultaneous inclusion of both work hours and schedule fit, we
estimated models in a progressive and logical order to ascertain that the addition of the theoretically relevant variables to the model did, in fact,
incrementally improve model fit. For the preliminary models (Models 1–3), the coefficients for
the predictors, their t-statistics, and the associated
p-values were not considered.
Our modeling strategy began by estimating a
single null model (Model 1) that contained only
the six outcomes described above. This model
yielded baseline variance estimates of the total
true score variation in each of the outcomes; these
were used in evaluating the additional proportion
of variance explained in each step. Specifically, as
predictors are added to the model, we expect unexplained variance to decline; the difference in
variance between the two models becomes the numerator in determining the proportion of variance
explained by the model, and the denominator is
the total variance from the null model. Although
there are six variances associated with each of the
six outcomes in our model, as compared with one
in a traditional regression model, the interpretation
1047
of explained variance (the difference in the unexplained variance in a model with more predictors subtracted from the unexplained variance in
a model with no predictors) over the original unexplained variance has an interpretation entirely
analogous to an R2 statistic; that is, the proportion
of variance explained by the model.
The next model (Model 2) added the control
variables (i.e., the number of children, household
income, marriage length, and each spouse’s negative affectivity). As control variables, these
would have been retained in subsequent models
regardless of the results of the hypothesis testing,
because we were interested in interpreting the effects of schedule fit net of the effect of these control variables.
We added to the resulting model the number
of hours worked by each member of the couple
to predict both their own and their partners’ outcomes (Model 3). As a conservative test of the
role of schedule fit, we retained the work hours
predictors regardless of the results of hypothesis
testing.
Finally, a single model estimating all six outcomes predicted by the control variables, the number of hours worked by each spouse, and husbands’ and wives’ assessments of two aspects of
schedule fit was estimated (Model 4). Hypothesis
testing was used to determine whether including
schedule fit incrementally improved the fit of the
model.
RESULTS
Model Estimation
Testing of the deviance statistic for the null (Model 1) and control (Model 2) models revealed that
adding control variables improved the fit of the
model (x2 5 141.98, df 5 24, p 5 .000). Further,
the control model explained variance in all three
outcomes, most notably psychological distress,
where approximately half the variance is explained by the control variables. The large reduction in the unexplained variance of psychological
distress when the control variables are added can
be accounted for by the addition of negative affectivity to the model. (See Brennan & Barnett,
1998, for a discussion of the role of negative affectivity in predicting psychological distress.)
Model 3 added hours worked to the control
model. The deviance test suggests that the null
hypothesis—no difference in the fit of the two
models—should not be rejected (x2 5 15.30, df
1048
Journal of Marriage and Family
TABLE 1. REGRESSION COEFFICIENTS ESTIMATION
FOR JOB-ROLE
QUALITY
Job-Role Quality
Wives
Predictor
Intercept
Self/self schedule fit
Partner/family schedule fit
Spouse self/self schedule fit
Spouse partner/family schedule fit
Work hours
Spouse work hours
Household income
Years married
Number of children
Negative affectivity
Unstandardized
Coefficient
(SE)
50.2037
3.1877*
2.8170
.8084
1.5396
.1033
2.2008*
2.8873
2.0478
.9851
2.4342**
Unexplained variance
Comparison model (Model 3)
Explanatory model (Model 4)
Additional variance explained (%)
Husbands
(.7601)
(1.3743)
(.8027)
(.6403)
(.9823)
(.0927)
(.0942)
(4.5404)
(.1277)
(1.1879)
(.1566)
62.00
47.89
14.1
Unstandardized
Coefficient
(SE)
49.4690
3.8974***
.8220
.0889
.3171
.0079
.0643
10.1427**
2.2295
2.0014
2.1456
(.7131)
(.7006)
(.8061)
(1.0023)
(.6724)
(.0815)
(.0862)
(3.7673)
(.1400)
(.9112)
(.1487)
64.78
41.17
23.6
Note: N 5 105 couples. For the explanatory model (Model 4), estimated parameters 5 88 and deviance statistic 5
7,634.21.
†p , .10. *p , .05. **p , .01. ***p , .001.
5 12, p 5 .225). In other words, the addition of
hours worked does not improve the fit of the model. The reduction in unexplained variation associated with the addition of hours worked to the
control model is generally small, although the unexplained variance in the husband’s marital-role
quality was reduced by about 4.9% when both
husbands’ and wives’ work hours were added to
the model.
Finally, a single model adding own and spouse
ratings of self/self and partner/family schedule fit
as predictors of each of the six simultaneous outcomes was estimated (Model 4). In this instance,
we reject the null hypothesis (x2 5 90.42, df 5
24, p 5 .000) of no improvement in model fit as
compared with Model 3 containing hours worked.
The addition of the schedule fit variables was associated with an across-the-board decrease in unexplained variance as compared with the model
containing only the control variables and work
hours, ranging from a very small reduction in the
unexplained variance of wives’ psychological distress (about 1.6%) to a considerably larger decrease in unexplained variance of husband’s jobrole quality (about 23.6%). Hence, we select the
full model (control variables, hours worked, and
schedule fit), Model 4, as the tested model that
best fits the data.
After fitting the models discussed above and
accepting Model 4 as our explanatory model, we
conducted several post hoc analyses to ensure that
certain findings, especially the finding that number
of hours worked cannot be associated within standard accepted levels of statistical confidence with
quality-of-life outcomes, were not the product of
multiple collinearity of predictor variables. Adding work hours to the null model (Model 1), the
control model (Model 2), and a model containing
the control variables plus schedule fit did not significantly improve model fit in any case. In contrast, adding schedule fit to the null model (Model
1), the control model (Model 2), and a model containing the control variables plus work hours
(Model 3) significantly improved model fit in every case. These findings suggest that number of
hours worked is not associated with quality-of-life
outcomes, and that this lack of association is not
due to multiple collinearity of predictor variables.
Hypothesis Testing
Job-role quality. As shown in Table 1, Hypothesis
1a was supported. That is, for women and men
alike, assessments of how well their work schedules were meeting their own needs predicted jobrole quality, with better self/self schedule fit predicting higher job-role quality. There was no
gender difference in the magnitude of the association between self/self schedule fit and job-role
quality (x2 5 0.379, p . .500). Hypothesis 1b was
Work Schedule Fit
1049
TABLE 2. REGRESSION COEFFICIENTS ESTIMATION
FOR
MARITAL-ROLE QUALITY
Marital-Role Quality
Wives
Predictor
Intercept
Self/self schedule fit
Partner/family schedule fit
Spouse self/self schedule fit
Spouse partner/family schedule fit
Work hours
Spouse work hours
Household income
Years married
Number of children
Negative affectivity
Husbands
Unstandardized
Coefficient
(SE)
Unstandardized
Coefficient
(SE)
49.8811
2.1598
2.2443*
22.7928**
1.8729†
.0578
.0049
212.3301*
.1883
2.5493
2.5180**
(.8323)
(1.1057)
(.9412)
(.8936)
(1.0263)
(.0827)
(.1087)
(5.7530)
(.1634)
(1.1486)
(.1913)
49.6709
21.3831
2.0569†
.9435
2.0329
.1615
2.0027
28.2069
2.2401
22.3367†
2.5854***
(.8947)
(.9277)
(1.0415)
(1.2447)
(.9549)
(.1172)
(.0950)
(6.2301)
(.2048)
(1.2532)
(.1336)
Unexplained variance
Comparison model (Model 3)
Explanatory model (Model 4)
Additional variance explained (%)
70.58
59.05
11.5
76.66
69.49
7.2
Note: N 5 105 couples. For the explanatory model (Model 4), estimated parameters 5 88 and deviance statistic 5
7,634.21.
†p , .10. *p , .05. **p , .01. ***p , .001.
not supported. As shown in Table 1, spouse ratings of partner/family schedule fit did not predict
job-role quality for men or for women.
Although only one of the job-role quality hypotheses was supported, the various components
of schedule fit together constitute a powerful explanatory construct. As shown at the bottom of
Table 1, adding own and spouse self/self and partner/family schedule fit to the model explained an
additional 14.1% of the variance in job-role quality for women and 23.6% of the variance in jobrole quality for men over and above the variance
explained by the controls plus work hours.
Interestingly, work hours did not predict jobrole quality for women or for men, either with or
without the schedule fit variables in the model.
There was, however, a significant crossover effect
among the reduced-hours physicians; specifically,
the more hours her husband worked, the poorer
her own job-role quality.
Marital-role quality. As shown in Table 2, Hypothesis 2a was partially supported. Among the
women, assessments of how well their partner’s
work schedule was meeting family needs predicted marital-role quality, with better partner/family
schedule fit predicting higher marital-role quality.
There was a similar trend (p 5 .051) among the
men. Although this effect was significant for
women and marginal for men, there was no gen-
der difference in the magnitude of the association
between partner/family schedule fit and maritalrole quality (x2 5 0.022, p . .500).
Hypothesis 2b concerning crossover effects
was partially supported. Among the women, there
was a trend (p 5 .071) for a crossover effect such
that the better her husband’s assessment of how
well her (reduced-hours) work schedule met family needs, the higher her marital-role quality. Although this trend was only seen among the wives,
the magnitude of the association between spouse’s
partner/family schedule fit rating and own maritalrole quality was not significantly different by gender (x2 5 2.331, p 5 .123).
There was also a significant crossover effect
among the women, although the direction was opposite to that predicted; specifically, the better her
husband’s assessment of how well his (full-time)
work schedule met his own needs, the poorer her
marital-role quality. There was no parallel finding
among the husbands, and this gender difference
was statistically significant (x2 5 8.613, p 5
.004).
As shown in Table 2, the components of schedule fit explained an additional 11.5% of the variance in marital-role quality for women and 7.2%
of the variance in marital-role quality for men
over and above the variance explained by the controls plus work hours.
1050
Journal of Marriage and Family
TABLE 3. REGRESSION COEFFICIENTS ESTIMATION
FOR
PSYCHOLOGICAL DISTRESS
Psychological Distress
Wives
Predictor
Intercept
Self/self schedule fit
Partner/family schedule fit
Spouse self/self schedule fit
Spouse partner/family schedule fit
Work hours
Spouse work hours
Household income
Years married
Number of children
Negative affectivity
Husbands
Unstandardized
Coefficient
(SE)
Unstandardized
Coefficient
(SE)
49.5677
2.5857
2.2502
.2949
21.0751
2.2157†
.1342
2.1102
2.0430
2.8542
1.0070***
(.7761)
(.9121)
(.6601)
(.7959)
(1.0010)
(.1148)
(.1058)
(5.0061)
(.1636)
(1.1279)
(.2300)
50.7316
2.6017
1.5800*
21.9936**
2.2847
.2991
2.1736†
210.1933*
.1887
2.8053
.9280***
(.6772)
(.7753)
(.6810)
(.7428)
(.7174)
(.0690)
(.0960)
(4.0839)
(.1171)
(.7905)
(.1982)
Unexplained variance
Comparison model (Model 3)
Explanatory model (Model 4)
Additional variance explained (%)
51.89
50.28
1.6
41.44
36.33
5.1
Note: N 5 105 couples. For the explanatory model (Model 4), estimated parameters 5 88 and deviance statistic 5
7,634.21.
†p , .10. *p , .05. **p .01. ***p , .001.
Psychological distress. As shown in Table 3, Hypothesis 3a was not supported. Among the women, there was no effect of partner/family schedule
fit on psychological distress, and among the men,
the finding was in the opposite direction to that
predicted. Specifically, husbands’ own ratings of
partner/family schedule fit predicted psychological distress. Hypothesis 3b was supported among
the husbands, but not among the wives. The
wives’ ratings of self/self schedule fit were negatively related to psychological distress among
their husbands, as predicted. Thus husbands’ psychological distress was higher when their wives
rated their own (reduced-hours) work schedule as
fitting their own needs poorly, but also when they
themselves rated their wives’ (reduced-hours)
work schedule as fitting family needs well. There
were no parallel findings among the wives, and
there was a trend for a gender difference in the
magnitude of the association between own partner/family schedule fit and own distress (x2 5
2.846, p 5 .088) and a significant gender difference in the magnitude of the association between
spouse ratings of self/self schedule fit and own
distress (x2 5 4.170, p 5 .039).
As shown in Table 3, there were also two marginally significant findings that are consistent with
the findings above. There were trends for both
husbands (p 5 .074) and wives (p 5 .063) to
report lower psychological distress when she
worked more hours. The table also shows that the
components of schedule fit explained an additional 1.6% of the variance in psychological distress
for women and an additional 5.1% of the variance
in distress for men over and above the variance
explained by the controls plus work hours. In the
case of the distress outcome in particular, it seems
likely that negative affectivity may be suppressing
some of the effects. The correlation between negative affectivity and distress is 0.646 (p 5 .000),
as contrasted with correlations of 20.311 (p 5
.000) and 20.307 (p 5 .000) between negative
affectivity and job-role quality and between negative affectivity and marital-role quality, respectively. Thus negative affectivity and distress share
41.7% of their variance, as opposed to a shared
variance of only 9.6% between negative affectivity and job-role quality and between negative affectivity and marital-role quality.
DISCUSSION
The main findings of this study of 105 women
Boston-area reduced-hours physicians and their
full-time–employed husbands are that self/self
schedule fit predicted higher job-role quality for
men and for women alike; that partner/family
schedule fit predicted higher marital-role quality
for women, with a similar trend among men; that
spouse’s rating of partner/family schedule fit
Work Schedule Fit
showed a trend to predict higher marital-role quality in a crossover effect among women; that
spouse’s self/self schedule fit predicted lower marital-role quality in a crossover effect among women; that partner/family schedule fit predicted higher psychological distress among men; and that
spouse’s self/self schedule fit predicted lower psychological distress in a crossover effect among
men.
Because half of the sample—the reduced-hours
physicians—had a restricted range of work hours,
all respondents were married, and most had children, the present analysis is a conservative test.
Employees who work more hours would probably
experience poorer schedule fit and worse outcomes, whereas unmarried and/or childless employees might have fewer nonwork commitments
and therefore experience better fit and outcomes.
There was still sufficient variability in the sample
for an adequate test of the hypotheses, however,
given that the women worked as many as 50 hours
per week and the men as many as 75 hours per
week. Further, with regard to the psychological
distress findings, given the degree of overlap that
exists between negative affectivity and distress,
the effects that we did find must have been robust
indeed in order to overcome the possible suppressing effect of negative affectivity.
The finding that self/self schedule fit predicts
higher job-role quality among both men and women is consistent with the results of previous research on schedule fit (Barnett et al., 1999) and
extends those findings to employees who work
longer hours than those in the original 1999 sample. In contrast, objective work characteristics
such as work hours per se do not predict job-role
quality, even among the men, whose work hours
were not restricted in the way that their wives’
work hours were. The fact that long work hours
do not appear to be associated with poor job-role
quality suggests that reduced-hours schedules per
se are not a panacea to the problem of professionals’ dissatisfaction with long work hours
(Boston Bar Association Task Force on Professional Challenges and Family Needs, 1999; Jacobs
& Gerson, 1997). Instead, a more productive strategy for addressing the time pressures of today’s
professionals may be to attempt to maximize their
self/self schedule fit, whether they work full-time
or reduced hours.
The findings that partner/family schedule fit
predicts marital-role quality among women, with
a similar trend among men, and that partner/family schedule fit predicts psychological distress
1051
among men highlight the importance of acknowledging that each family member’s social-role experiences affect, and are in turn affected by, other
family members’ social-role experiences. There is
also an interesting trend, at least among the women, to rate their own marital-role quality higher
when their husbands assessed their work schedules as fitting family needs well. This finding,
along with the crossover effects found in the present study, further illustrates the interdependence
of husbands’ and wives’ experiences.
Some of our unexpected findings probably result from the specific composition of the present
sample, in which all of the wives were physicians
who had reduced their work schedules, whereas
their husbands, the majority of whom were also
physicians, continued to work full-time. Thus,
these women had traded off time at work in exchange for more family time, whereas their husbands had not. Given the prevailing culture in the
medical profession of long hours and total commitment to work, such a tradeoff may be experienced as a career sacrifice, and one which may
involve actual penalties in terms of salary and advancement. In this context, it is easier to understand why husbands’ satisfaction with their fulltime schedules might be negatively related to
wives’ marital-role quality. We do have some evidence that a sizable number of the wives perceived their reduced-hours schedules as a career
sacrifice. Some 39.1% reported that having to forgo some aspects of their work that they valued in
exchange for their reduced schedule was of considerable or extreme concern to them, and 31.4%
reported the same level of concern about being
less advanced professionally than others with
whom they had trained.
It could also be argued that in addition to the
wives sacrificing work time in exchange for more
family time, their husbands are sacrificing family
time in exchange for more work time, and that to
the extent that the husband’s unhappiness with his
schedule is an indicator that he perceives his
heavy work schedule to be a sacrifice, his wife
reports higher marital-role quality. The interpretation that husbands are making a sacrifice is supported by the fact that more husbands than wives
reported working too many hours (57.1% vs.
39.0%, respectively, as previously discussed in the
Sample section). It does not appear to be the case,
however, that the husbands are working more
hours in order to allow their wives to work fewer
hours. First, as reported in the Sample section, the
husbands had been working their schedules for 3
1052
years longer, on average, than their wives had,
suggesting that the husbands’ work schedules did
not change at the time that their wives reduced
their own work hours. Second, the majority of the
husbands in the present sample were also physicians, and their work schedules are typical of
those of other full-time physicians, rather than
representing increased hours. In any case, future
research should examine in greater detail the relative sacrifices of husbands and wives in dualearner couples with one member working reduced
hours.
Other results suggest that the men in the present sample did appear to be sensitive to their
wives’ experiences, as evidenced by their high
psychological distress when their wives were unhappy with their reduced-hours schedules. The
men’s seemingly paradoxical distress when they
themselves believed that their wives’ work schedules were meeting family needs may also reflect
an awareness that their wives may be paying a
career price in order to spend more time taking
care of family. These results are compatible with
the trends for both husbands and wives to report
lower psychological distress when the wife is
working longer hours, as if both are happier when
she makes a smaller adjustment to her work
schedule.
The findings discussed above are consistent
with the findings of Brockwood et al. (2002) that
work adjustments can have negative as well as
positive effects on women, and with Carr’s (2002)
argument that career-invested women may pay a
price when they trade off work for family. Our
findings extend Carr’s argument by suggesting
that there may be negative effects for these women’s husbands as well. Poor outcomes may be especially likely in the context of an asymmetrical
arrangement in which only one spouse and not the
other makes a work adjustment to meet family
needs. According to equity theory (Walster, Walster, & Berscheid, 1978), both members of such
couples might experience poor outcomes, because
inequitable relationships are distressing for the overadvantaged as well as for the underadvantaged
partner.
Although different populations might show different patterns of relationships between schedule
fit and outcomes, we expect that the findings linking self/self schedule fit to job-role quality would
generalize to other employed populations and that
the findings linking partner/family schedule fit to
marital-role quality would generalize to other
married populations in which at least one spouse
Journal of Marriage and Family
is employed. We argue that even the unexpected
findings discussed above might generalize to any
dual-professional couple in which one member,
whether it be the wife or the husband, has made
a temporary or permanent work adjustment to
meet family needs. Dual-professional couples are
becoming increasingly common as women continue to achieve high-level positions in the professions; therefore, it is important to further delineate
the positive and negative effects on professionals
of making such work adjustments. Further research should also address the question of whether
these effects are gender-specific, or whether they
would apply equally to men and to women who
accommodate their work to their family needs.
In addition to the generalizability issues, another limitation to this study is that the data are
cross-sectional. It would be useful to collect longitudinal data, starting before any work adjustments are made and continuing until the employee
reverses the adjustment (e.g., goes back to a fulltime schedule), if applicable. Such data could illuminate questions such as whether outcomes depend on how the initial decision was made, on the
duration of the adjustment, on whether one or
both partners make adjustments, on the perceived
career costs of the adjustment, and so on.
Another topic for further research is additional
delineation of the ways in which various aspects
of work schedules mesh with particular family
needs. An ongoing project involves the assessment of a new measure of subjective experience
of work schedule that includes specific rewards
and concerns about the effects of different work
schedules on oneself and on other family members; these rewards and concerns were identified
in open-ended interviews with women who were
day- and evening-shift nurses and their husbands.
Despite the limitations discussed above, the
present study makes several contributions. First,
our findings highlight the value of subjective indicators of work experiences such as schedule fit.
This study provides additional evidence of the
utility of the schedule fit construct in predicting
outcomes, as well as further delineating the components of the schedule fit measure into self/self
and partner/family aspects of schedule fit. Our
findings also further illuminate the tradeoffs that
may be involved when people make adjustments
to their work schedules in order to better manage
work and family demands. In addition, by fully
exploiting the couple nature of our data and by
accommodating the inherent relatedness of the experiences of couple members, we are able to dem-
Work Schedule Fit
1053
onstrate the interdependence of spouses’ work-family experiences. This interdependence reinforces
the importance of treating the couple as the unit
of analysis in future work-family research. In sum,
the present study contributes to understanding the
larger ecology of work and family life by shedding light on some of the specific linkages between dual-earner husbands’ and wives’ work and
family experiences.
NOTE
Data for this analysis were gathered under a grant from
the Alfred P. Sloan Foundation to the second author
(R.C.B.) and Lena Lundgren, PhD, at Boston University. We gratefully acknowledge the contributions of our
interviewers, Joyce Buni, Maureen Crowley, Martie Dumas, Connie Festo, Carla Fink, Phyllis Jansen, Rose
Marie Pronovost, Marilyn Spiegel, and Jayne Wilson.
REFERENCES
Bailyn, L. (1993). Breaking the mold: Women, men, and
time in the new corporate world. New York: Free
Press.
Barling, J. (1984). Effects of husbands’ work experiences on wives’ marital satisfaction. Journal of Social
Psychology, 124, 219–225.
Barnett, R. C. (1998). Toward a review and reconceptualization of the work/family literature. Genetic, Social and General Psychology Monographs, 124, 125–
182.
Barnett, R. C., & Brennan, R. T. (1995). The relationship between job experiences and psychological distress: A structural equation approach. Journal of Organizational Behavior, 16, 259–276.
Barnett, R. C., & Brennan, R. T. (1997). Change in job
conditions, change in psychological distress, and gender: A longitudinal study of dual-earner couples.
Journal of Organizational Behavior, 18, 253–274.
Barnett, R. C., & Brennan, R. T. (1998). Change in job
conditions and change in psychological distress within couples: A study of crossover effects. Women’s
Health: Research on Gender, Behavior, and Policy,
4, 313–339.
Barnett, R. C., Brennan, R. T., Raudenbush, S. W.,
Pleck, J. H., & Marshall, N. L. (1995). Change in job
and marital experiences and change in psychological
distress: A longitudinal study of dual-earner couples.
Journal of Personality and Social Psychology, 69,
839–850.
Barnett, R. C., Gareis, K. C., & Brennan, R. T. (1999).
Fit as a mediator of the relationship between work
hours and burnout. Journal of Occupational Health
Psychology, 4, 307–317.
Barnett, R. C., & Hyde, J. S. (2001). Women, men,
work, and family: An expansionist theory. American
Psychologist, 56, 1–16.
Barnett, R. C., Marshall, N. L., Raudenbush, S., &
Brennan, R. (1993). Gender and the relationship between job experiences and psychological distress: A
study of dual-earner couples. Journal of Personality
and Social Psychology, 64, 794–806.
Blau, F. D., Ferber, M. A., & Winkler, A. E. (1998). The
economics of women, men and work (3rd ed.). Upper
Saddle River, NJ: Prentice Hall.
Boston Bar Association Task Force on Professional
Challenges and Family Needs. (1999). Facing the
grail: Confronting the cost of work-family imbalance.
Boston, MA: Boston Bar Association.
Bowen, G. L. (1998). Effects of leader support in the
work unit on the relationship between work spillover
and family adaptation. Journal of Family and Economic Issues, 19, 25–52.
Bowen, G. L., Orthner, D. K., & Bell, D. B. (1997).
Differences in spouses’ perceptions of family adaptation. Journal of Social Behavior and Personality,
12, 53–72.
Bray, J. H. (1995). Methodological advances in family
psychology research: Introduction to the special section. Journal of Family Psychology, 9, 107–109.
Bray, J. H., Maxwell, S. E., & Cole, D. (1995). Multivariate statistics for family psychology research.
Journal of Family Psychology, 9, 144–160.
Brennan, R. T., & Barnett, R. C. (1998). Negative affectivity: How serious a threat to self-report studies
of psychological distress? Women’s Health: Research
on Gender, Behavior, and Policy, 4, 369–384.
Brennan, R. T., Barnett, R. C., & Gareis, K. C. (2001).
When she earns more than he does: A longitudinal
study of dual-earner couples. Journal of Marriage
and Family, 63, 168–182.
Brennan, R. T., Kim, J., Wenz-Gross, M., & Siperstein,
G. (2001). The relative equitability of high-stakes
testing versus teacher-assigned grades: An analysis of
the Massachusetts Comprehensive Assessment System (MCAS). Harvard Educational Review, 7, 173–
216.
Brockwood, K. J., Hammer, L. B., Neal, M. B., & Colton, C. L. (2002). Effects of accommodations made
at home and at work on wives’ and husbands’ family
and job satisfaction. Journal of Feminist Family
Therapy, 13, 41–64.
Burke, M. J., Brief, A. P., & George, J. M. (1993). The
role of negative affectivity in understanding relations
between self-reports of stressors and strains: A comment on the applied psychology literature. Journal of
Applied Psychology, 78, 402–412.
Carr, D. (2002). The psychological consequences of
work-family trade-offs for three cohorts of men and
women. Social Psychology Quarterly, 65, 103–124.
Crouter, A. C., Bumpus, M. F., Head, M. R., & McHale,
S. M. (2001). Implications of overwork and overload
for the quality of men’s family relationships. Journal
of Marriage and Family, 63, 404–416.
Derogatis, L. R. (1975). The SCL-90-R. Baltimore: Clinical Psychometrics.
Gutek, B. A., Searle, S., & Klepa, L. (1991). Rational
versus gender role explanations for work-family conflict. Journal of Applied Psychology, 76, 560–568.
Hammer, L. B., Allen, E., & Grigsby, T. D. (1997).
Work-family conflict in dual-earner couples: Within
individual and crossover effects of work and family.
Journal of Vocational Behavior, 50, 185–203.
Hughes, D., & Galinsky, E. (1994). Gender, job and
family conditions and psychological symptoms. Psychology of Women Quarterly, 18, 251–270.
Hyde, J. S., & Plant, E. A. (1996). Factor structure of
Barnett’s partner rewards and concerns scale. Madison: University of Wisconsin.
1054
Jacobs, J. A., & Gerson, K. (1997). The endless day or
flexible office? Working hours, work-family conflict,
and gender equity in the modern workplace. Report
to the Alfred P. Sloan Foundation, New York.
MacDermid, S. M., Barnett, R., Crosby, F., Greenhaus,
J., Koblenz, M., Marks, S., Perry-Jenkins, M., Voydanoff, P., Wethington, E., & Sabbatini-Bunch, L.
(2000). The measurement of work/life tension: Recommendations of a virtual think tank. Boston, MA:
Alfred P. Sloan Foundation.
Maguire, M. C. (1999). Treating the dyad as the unit of
analysis: A primer on three analytic approaches.
Journal of Marriage and the Family, 61, 213–223.
Marks, S. R., Huston, T. L., Johnson, E. M., & MacDermid, S. M. (2001). Role balance among white
married couples. Journal of Marriage and Family,
63, 1083–1098.
Marks, S. R., & MacDermid, S. M. (1996). Multiple
roles and the self: A theory of role balance. Journal
of Marriage and the Family, 58, 417–432.
Milkie, M. A., & Peltola, P. (1999). Playing all the roles:
Gender and the work-family balancing act. Journal
of Marriage and the Family, 61, 476–490.
Moen, P., & Wethington, E. (1992). The concept of family adaptive strategies. Annual Review of Sociology,
18, 233–251.
Ozer, E. M., Barnett, R. C., Brennan, R. T., & Sperling,
J. (1998). Does child care involvement increase or
decrease distress among dual-earner couples? Women’s Health: Research on Gender, Behavior, and Policy, 4, 285–311.
Parsons, J. A., Warnecke, R. B., Czaja, R. F., Barnsley,
J., & Kaluzny, A. (1994). Factors associated with response rates in a national survey of primary care physicians. Evaluation Review, 18, 756–766.
Pittman, J. F. (1994). Work/family fit as a mediator of
work factors on marital tension: Evidence from the
interface of greedy institutions. Human Relations, 47,
183–209.
Raudenbush, S. W., Brennan, R. T., & Barnett, R. C.
(1995). A multivariate hierarchical model for study-
Journal of Marriage and Family
ing psychological change within married couples.
Journal of Family Psychology, 9, 161–174.
Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical
linear models: Applications and data analysis methods. Thousand Oaks, CA: Sage.
Raudenbush, S. W., Bryk, A. S., Cheong, Y. F., & Congdon, R. T. (2000). HLM 5: Hierarchical linear and
nonlinear modeling. Lincolnwood, IL: Scientific
Software International.
Rook, K., Dooley, D., & Catalano, R. (1991). Stress
transmission: The effects of husbands’ job stressors
on the emotional health of their wives. Journal of
Marriage and the Family, 53, 165–177.
Sobal, J., DeForge, B. R., Ferentz, K. S., Muncie, H.
L., Valente, C. M., & Levine, D. M. (1990). Physician
responses to multiple questionnaire mailings. Evaluation Review, 18, 711–722.
Spain, D., & Bianchi, S. (1996). Balancing act: Motherhood, marriage, and employment among American
women. New York: Sage.
Spielberger, C. D. (1983). Manual for the State-Trait
Anxiety Inventory (Form Y ed.). Palo Alto: Consulting Psychologists Press.
Supovitz, J. A., & Brennan, R. T. (1997). Mirror, mirror
on the wall, which is the fairest test of all? An examination of the equitability of portfolio assessment
relative to standardized tests. Harvard Educational
Review, 67, 472–506.
Walster, E., Walster, G. W., & Berscheid, E. (1978). Equity: Theory and research. Boston: Allyn & Bacon.
Westman, M., & Etzion, D. (1995). Crossover of stress,
strain and resources from one spouse to another. Journal of Organizational Behavior, 16, 169–181.
Wethington, E., & Kessler, R. C. (1989). Employment,
parental responsibility, and psychological distress: A
longitudinal study of married women. Journal of
Family Issues, 10, 527–546.
Windle, M., & Dumenci, L. (1997). Parental and occupational stress as predictors of depressive symptoms among dual-income couples: A multilevel modeling approach. Journal of Marriage and the Family,
59, 625–634.